Modulation of the Regulatory Activity of Bacterial Two-component Systems by SlyA*

Activation of the transcriptional regulator SlyA by the PhoP/PhoQ two-component system controls intracellular expression of numerous factors influencing Salmonella virulence. By dissecting the SlyA regulon using stable isotope labeling with amino acids in cell culture analysis, we found that SlyA enhances overall transcription of PhoP-activated loci. This amplification of cellular responses to Mg2+ occurs when SlyA binds to the phoPQ promoter thereby activating phoP autoregulation via a positive feedback mechanism. SlyA footprints a DNA region located one helical turn upstream of the PhoP box, which overlaps the H-NS-binding motif required for signal-dependent phoP repression in high Mg2+ conditions. Therefore, binding of SlyA likely antagonizes H-NS and facilitates the interaction of PhoP to its own promoter, subsequently activating the phoPQ operon. Establishment of this regulatory circuit allows SlyA to exert its effect on the PhoP/PhoQ system specifically in Salmonella, which may confer an additional transcriptional regulation. Thus, our results provide a molecular mechanism that determines SlyA-dependent activation of PhoP-regulated genes in modulating Salmonella virulence. Evidence from this study also suggests a function of SlyA as a mediator in signal transduction from the PhoP/PhoQ system to other bacterial two-component systems in Salmonella.

The ability to sense hostile environments and trigger compensatory gene expression is critical for Salmonella typhimurium to survive within host cells (review see Refs. 1, 2). The PhoP/PhoQ two-component system governs regulatory signaling networks by responding to environmental changes, including acidic pH, low Mg 2ϩ , and host-derived antimicrobial peptides, which confer bacterial resistance to depletion of Mg 2ϩ as well as bactericidal substances (3)(4)(5). When bacteria are grown under inducing conditions, kinase activity of the sensor in a two-component system is modulated to transfer a phosphoryl group from ATP to its cognate response regulator, which enhances the modified regulator to interact with its target promoters (for review see Ref. 6). Accordingly, PhoQ mediates phosphorylation of PhoP to facilitate binding of this regulator to the "PhoP box" sequence in promoter regions, thus giving rise to gene regulation (7,8).
The PhoP/PhoQ system functions as a master regulatory system that controls expression of various transcriptional regulators such as the two-component systems PmrA/PmrB (9) and RstA/RstB (10), as well as the MarR family member SlyA (11,12). SlyA, which is present in members of the family Enterobacteriaceae, has been recognized as a transcriptional regulator specifically modulating the intracellular expression of chromosomal loci required for Salmonella growth in macrophages (13) and resistance of internalized bacteria to oxidative stress (14). The PhoP/PhoQ system may contribute to Salmonella virulence, in part, by regulating slyA expression because observations from different laboratories suggest that SlyA is involved in regulation of a subgroup of PhoP-dependent genes (12,15,16). However, SlyA does not simply function as an intermediate regulator for the PhoPdirected regulation, but rather it includes more complicated regulatory circuits impinging on PhoP activity itself. Several previous results showed that a feedforward regulatory loop directs the expression of ugtL and pagC, whose promoter regions possess binding sites for both PhoP and SlyA (12,16). Recently, we demonstrated that binding of PhoP and SlyA is to antagonize the inhibitory activity of the transcriptional repressor H-NS in these promoters, which occupies both the PhoP and SlyA boxes in signal-depleting conditions (17). As a result, transcription of ugtL and pagC is greatly activated only when both regulators are simultaneously present (12,16). On the other hand, it remains to be determined whether SlyA could exert a direct regulatory effect on other PhoPand SlyA-dependent loci identified from a transcriptomic analysis (16).
Transcription of the phoPQ operon is positively autoregulated (18). This regulation is also responsible for transcriptional expression in other two-component systems, e.g. ompR/envZ (19). Regardless of the important role played by the PhoP/PhoQ system in bacterial virulence, it is not clear whether other regulatory mechanisms are involved in controlling expression of the phoPQ operon in Salmonella. Here we demonstrate that SlyA fine-tunes the cellular level of the PhoP/PhoQ system. Our results provide evidence that SlyA participates in a positive feedback loop, which facilitates transcription of the phoPQ loci and, in turn, stimulates transcription of the PhoP regulon. Our data also suggest that SlyA functions as a connecting mediator that transmits signals from the PhoP/PhoQ system to several other two-component systems in Salmonella.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Strains used in this study are described in Table 1. All Salmonella enterica serovar Typhimurium strains are derived from the wild-type strain 14028s. Escherichia coli was derived from the wild-type strain MC4100. Bacteria were grown at 37°C in Luria-Bertani broth or in N minimal medium (20) (pH 7.4) supplemented with 38 mM glycerol and 0.1% casamino acids, except SILAC 3 experiments. MgCl 2 is added to required concentrations. When necessary, antibiotics are added at final concentrations of 50 g/ml for ampicillin, 20 g/ml for chloramphenicol, 50 g/ml for kanamycin, and 12.5 g/ml for tetracycline. E. coli DH5␣ is used as a host for the preparation of plasmid DNA.
Construction of Chromosomal Mutations, lac Fusions, and Epitope-tagged Proteins-Oligonucleotides used as probes for the construction of strains and plasmids are described in Table  2. PCR products were used to generate coding region deletions, introduce FLAG/HA epitope sequence, or introduce a scar sequence for the lacZ fusion in bacterial chromosome as described previously (21). Primers were listed as pairs for individual genes in Table 2, and plasmid pKD3 was used as a template. All resulting strains were confirmed using colony PCR and DNA sequencing. A lac gene was integrated behind a coding region, in chromosome, using plasmid pKG137, into the FLP recombination target sequence generated after the Cm R cassette was removed using plasmid pCP20 (22). The up-scar-phoP-HA strain (YS11591) was constructed as follows. Cm R cassette was introduced upstream of the phoP promoter using a PCR fragment synthesized with primers 395 and 396 from pKD3. DNA amplification was then carried out using chromosomal DNA from the above Cm R strain as template and primers 28 and 395. This DNA product was electroporated into wild type harboring pKD46, and Cm R colonies were selected. The HA fusion was confirmed using colony PCR and DNA sequencing. The Cm R cassette was removed using plasmid pCP20 (22). Phage P1-mediated transductions in E. coli and phage P22mediated transductions in Salmonella were performed as described previously (23,24).
Construction of Plasmids-Plasmids pYS1109 and pYS1177 were constructed by digesting PCR fragments containing 472 bp of the wild-type slyA gene, generated with primers 45 and 46, and 675 bp of the wild-type phoP gene, generated with primers 88 and 89, and strain 14028s chromosomal DNA as template, with BamHI and HindIII and cloning between the BamHI and HindIII sites of pUHE21. Plasmid pYS1055 was constructed by digesting a PCR fragment containing 437 bp of the wild-type slyA gene, generated with primers 478 and 479 and strain 14028s chromosomal DNA as template, with EcoRI and BamHI and cloning between the EcoRI and BamHI sites of pET11a. Inserted DNA sequences in plasmids were confirmed by DNA sequencing.
Analysis of slyA Gene Profile Using SILAC-We conducted SILAC analysis using the following steps modified from an original reference (25). (i) For preparation of bacterial samples, the growth conditions were 37°C for 4 h aerobically in N minimal medium (pH 7.4; no casamino acids added) supplemented with 0.01 mM MgCl 2 and 40 g/ml L-[ 13 C 6 ]arginine and L-[ 13 C 6 ]lysine (Cambridge Isotope Laboratories, Inc.) for wild type or 40 g/ml normal L-arginine and L-lysine (Sigma) for the slyA mutant. Salmonella cells were collected and opened, and the supernatant was separated from the cell debris. Samples from the wild-type strain and the slyA mutant with equal (vii) For peptide extraction from gel after trypsin digestion, the suspension from trypsin treatment was transferred to a new Eppendorf tube, and a 200-l solution of 5% trifluoroacetic acid, 50% acetonitrile was added to the tube with gel pieces and incubated with gentle oscillation at room temperature for 60 min. The supernatant was combined with previous fractions. This extraction was repeated again by adding another 200 l of 5% trifluoroacetic acid, 50% acetonitrile to the gel pieces for 60 min; the supernatant was then combined with the above fractions. The volume of supernatant was reduced to 30 l using Speed vacuuming at room temperature, and then an equal amount (30 l) of buffer containing 3% acetonitrile and 0.1% formic acid in water was added. (viii) For peptide high pressure liquid chromatography separation and liquid chromatography/ electrospray ionization/MS/MS mass spectrometer analyses, samples (0.2-1.4 l) were injected into a Micromass CapLC liquid chromatography system (Micromass, Manchester, UK) and concentrated in a PepMap C18 precolumn (300 m ϫ 5 mm). The precolumn was washed (3 min, 0.1% formic acid, flow rate of 30 l/min), and then the peptide mixture was eluted into an analytical C18 column (150 mm ϫ17 m) and analyzed using a solvent gradient from solution A (3% acetonitrile) to solution B (95% acetonitrile) containing 0.1% formic acid over 50 min at flow rates gradually reduced from 5 l/min to 200 nl/min by stream splitting. Liquid chromatography eluent was applied into the nanoflow source of a Q-TOF micromass spectrometer (Micromass, Manchester, UK). Sample running conditions were set as follows: the source temperature was 80°C; and the cone gas flow was 50 liters/h. A voltage of 3.2 kV was applied to the nanoflow probe tip, and data were acquired in positive ion mode. Survey scans were integrated over 1 s, and MS/MS scans were integrated over 3 s. Switching from survey to MS/MS scan mode was performed in a data-dependent manner. The maximum MS/MS to survey scan ratio was three. Collision energy was 28 eV. Data were processed with Masslynx 3.5 software. Multipoint calibration was performed using selected fragment ions produced by CAD of Glu-fibrinopeptide B. MS/MS spectra were processed by Masslynx software to generate a peak list file as described previously (26).
Reverse Transcription-PCR-Bacterial cells were grown for 4 h in N medium supplemented with 0.01 mM and 10 mM MgCl 2 . Expression of the slyA gene and the phoP gene was induced from strains harboring pYS1109 and pYS1177 by adding 0.2 mM IPTG under the same growth conditions. Total RNA was isolated from bacterial culture using SV Total RNA Isolation System (Promega) according to the manufacturer's instructions. RNA concentration was determined by spectrophotometry at 260 nm. RNA quality was confirmed by agarose gel electrophoresis. cDNA was synthesized using murine leu-kemia virus reverse transcriptase and random primers (New England Biolabs). DNA was amplified with primers listed in Table 2 using Taq polymerase (New England Biolabs) and performed in a thermocycler (Bio-Rad). Quantification was conducted using software Quantity One (Bio-Rad).
Immunoblot Analysis of Epitopetagged Proteins-Strains harboring encoded proteins with a C-terminal HA or FLAG epitope were grown in 25 ml of N medium as described above for 4 h, washed with PBS once, resuspended in 0.5 ml of PBS, and opened by sonication. Expression of the slyA gene and the phoP gene was induced from strains harboring pYS1109 and pYS1177 by adding 0.2 mM IPTG under the same growth conditions. The tagged proteins from whole-cell lysates were separated in 12.5% SDS-polyacrylamide gels and detected using the immunoblot analysis (ECL, Pierce). Quantification was conducted using software Quantity One (Bio-Rad).
␤-Galactosidase Assays-Galactosidase assays were carried out in triplicate, and the activity was determined as described previously (23). Data correspond to three independent assays conducted in duplicate. Percentage of ␤-galactosidase activity in Fig. 5B was calculated by the following: (␤-galactosidase activity in y mM Ϭ ␤-galactosidase activity in 0.01 mM) ϫ 100. y mM means a given Mg 2ϩ concentration we tested. C1 ⁄ 2 represents the experimentally determined value of Mg 2ϩ concentration, which allows percentage of ␤-galactosidase activity to reach 50% maximum level.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assays were performed as described previously (27). 0.2 mM IPTG was supplemented in N medium for pYS1109-carrying and pYS1177-carrying bacteria. The phoP promoter region was detected by PCR using primers 409 and 410 (Table 2).
Electrophoretic Mobility Shift Assay (EMSA)-1 pmol of 32 Plabeled phoP DNA fragment amplified with primers in Table 2 was incubated at room temperature for 30 min with 50 pmol of SlyA-FLAG protein in 15 l of an EMSA buffer consisting of 10 mM Tris-HCl (pH 8.0), 50 mM NaCl, 50 mM KCl, and 10 mM MgCl 2 . The "cold" DNA was added in excess of the "hot DNA" amount when required. Monoclonal anti-FLAG M2 (ϳ1 g, Sigma), as well as FLAG octapeptide (Sigma), was used when required. After the addition of the DNA dye solution (40% glyc-  OCTOBER 17, 2008 • VOLUME 283 • NUMBER 42 erol, 0.05% bromphenol blue, 0.05% xylene cyanol), the mixture was directly subjected to 5% PAGE. Signals were detected by autoradiography.

SlyA Feedback Activates the PhoP/PhoQ Two-component System
DNase I Protection Assays-DNase I protection assays were carried out using DNA fragments amplified by PCR using Salmonella chromosomal DNA as template. Prior to the PCR, primers 469 and 470 (Table 2) were labeled with T4 polynucleotide kinase and [␥-32 P]ATP. The phoP promoter region was amplified with primers 32 P-labeled 469 and 470 for the coding strand or with 469 and 32 P-labeled 470 for the noncoding strand. Approximately 25 pmol of labeled DNA and 0, 50, 100, or 200 pmol of the SlyA-FLAG protein were used in a 100-l reaction. DNase I digestion was carried out as described previously (8). DNase I was purchased from Invitrogen, and 0.05 units were used per reaction. Samples (3 l) were analyzed by 6% denaturing PAGE by comparison with a DNA sequence ladder generated with the appropriate primer by using a Maxam and Gilbert AϩG reaction. The positions of radioactive DNA fragments in the gels were detected by autoradiography.
Isolation of slyA-FLAG-E. coli BL21 (DE3) harboring plasmid pET11a-slyA-FLAG (pYS1055) was grown at 37°C and shaking to A 600 nm 0.5 in 500 ml of LB medium; then IPTG (final concentration, 1 mM) was added, and bacteria were incubated for another 2 h. Cells were harvested, washed with PBS once, resuspended in 10 ml of PBS, and opened by sonication. The whole-cell lysate was used for SlyA-FLAG purification by mixing with EZview Red Anti-FLAG M2 affinity Gel (Sigma) following the instructions from the manufacturer. Pure SlyA-FLAG sample was tested using silver staining (Pierce) following the instructions from the manufacturer.

Enhanced Expression of the PhoP/PhoQ System by SlyA-A
cDNA microarray showed that SlyA influences many PhoPregulated loci (16). However, a SlyA-FLAG fusion protein was unable to interact with every promoter of these genes because not all intergenic regions can be gel-shifted by this protein (data not shown). This observation suggests that SlyA should regulate these genes indirectly by modulating regulatory activity of another transcriptional regulator. To find this possible factor not yet identified in the previous studies, we employed a new approach by conducting a quantitative proteomic analysis (i.e. SILAC, see Ref. 25) to profile protein levels affected by SlyA in Salmonella under a growth condition in which transcription of the slyA gene is activated (11,12). The experiment design is illustrated in Fig. 1A, in which wild-type culture was supplemented with stable isotope 13 C-labeled arginine and lysine (shown as Arg-13 C 6 and Lys-13 C 6 , respectively), whereas the slyA mutant was supplemented with normal arginine and lysine (Arg-12 C 6 and Lys-12 C 6 , respectively). Comparison of a peptide synthesized in wild type and the slyA mutant from constitutively expressed genes yields a ratio of 13 C-labeled peptide versus normal 12 C-embedded peptide (shown as 13C/12C in this work) ϳ0.7 (results of the arcA and ompA gene products in Table 3, and data not shown). This is because Salmonella possesses de novo synthesis pathways for arginine and lysine (for review see Ref. 28); therefore, both normal 12 C-embedded and exogenous 13 C-labeled amino acids will be incorporated into proteins of wild-type cells. Genes that produced proteins with ratios above 0.7 were regarded as SlyA-activated loci. Consistent with this definition, five peptides from SlyA-activated PagC (16) gave an average 13C/12C ratio of 13.39 (Table 3). In this study, only selected chromosomal loci responding to SlyA regulation are listed in Table 3 from our ongoing SILAC analysis.
Three peptides derived from PhoP have an average 13C/12C ratio of 3.34 (Table 3), indicating that this regulator should be up-regulated by SlyA. Consequently, SlyA may facilitate the overall expression of PhoP-activated genes by raising the level of the PhoP/PhoQ system. Indeed, results from SILAC in Table  3 revealed that SlyA activates several loci dependent directly or indirectly on PhoP (10,29,30), including mgtB (average 13C/ 12C ϭ 3.00), ugd (2.90), pmrF (1.43), yaiB (2.41), pmrA (2.50), and rstA (1.66). However, a gel-shift result showed that the SlyA-FLAG protein could not interact with promoter regions of the mgtCB and yaiB loci (data not shown). Taken together, these results demonstrate that SlyA enhances the expression of PhoP-activated genes by a feedback up-regulation of the PhoP/ PhoQ system.
In addition, SILAC results show that several two-component systems are modulated by SlyA (Table 3), indicating that SlyA might enhance expression of these systems that were not yet associated with the PhoP/PhoQ system. The OmpR/EnvZ system (31) appears to be activated by SlyA because the average 13C/12C ratio of two peptides from regulator OmpR was 3.76. Meanwhile, HtrA, a protease activated by the CpxR/CpxA sys- tem (32), was up-regulated by SlyA (13C/12C ϭ 1.27). However, the SlyA-mediated activation is not applicable to every two-component system in Salmonella. This realization stemmed from our observation that the ArcA regulator protein of the ArcA/ArcB system had a 13C/12C ratio of 0.7 (Table 3).
Immunoblot analyses were performed to evaluate the SILAC results using strains harboring HA or FLAG epitope immediately upstream of the stop codon of their chromosomal loci for two-component regulator genes and other genes. Consistent with previous results (16), the PagC protein level was dramatically reduced in a slyA mutant when compared with wild type (Fig. 1B). Meanwhile, we found that protein levels of PhoP, OmpR, CpxR, PmrA, and MgtB were lower in slyA mutants than those levels in wild-type strains, whereas protein levels of ArcA and the control CorA were similar in wild-type and the slyA mutant (Fig. 1B). Peptides representing components from other two-component systems were unable to be identified in the current SILAC assays. One possible reason was that synthesized proteins in the current growth condition were below detectable levels.
SlyA Facilitates Transcription of the phoPQ Operon-SlyA seems unlikely to control a post-transcriptional modification of PhoP because the level of this protein became similar when it was expressed from heterologous promoter P lac (Fig. 2A). To see if SlyA could exert its effect directly on phoP transcription, we determined mRNA levels of the phoP transcript corresponding to the first 100 nucleotides of the coding region. When bacterial cells were grown in 0.01 mM Mg 2ϩ , the phoP mRNA level was ϳ2.5-fold higher in wild type than the isogenic slyA mutant (Fig. 2B). Consistently, the PhoP protein level was ϳ3-fold higher in wild type than the slyA mutant (Fig. 2B). SlyA does not influence overall mRNA levels because the mRNA level of a constitutive gene, rpoD, was similar in both wild type and slyA mutant, nor does it influence overall protein levels because the level of CorA protein was similar in both strains (Fig. 2B). The reduced phoP mRNA level or PhoP protein level in the slyA mutant could be recovered by a plasmid (pYS1109, pslyA) carrying a wild-type copy of the slyA-FLAG fusion, indicating that this phenotype resulted solely from an absence of the SlyA protein that was also demonstrated by a Western blot analysis (Fig.   FIGURE 2. SlyA facilitates transcriptional activity of the PhoP/PhoQ system in Salmonella. Bacteria were grown for 4 h in N medium (pH 7.4) containing 0.01 mM Mg 2ϩ in the following assays. 0.2 mM IPTG was supplemented to growth of the bacterial cells harboring pUHE21-2lac q with a slyA-FLAG fusion (pYS1109, pslyA) or pUHE21-2lac q with a phoP-HA fusion (pYS1177, pphoP). A, Western blot analysis of cell extracts prepared from wild type (14028s) and the slyA mutant (YS11068) harboring plasmid pYS1177. Monoclonal anti-HA antibodies (Sigma) were used. Percentage of the protein (Pr) amount was calculated by formula: % Pr ϭ (density of test strain Ϭ density of wild type) ϫ 100. B, mRNA levels of phoP were determined using RT-PCR analysis in wild type (14028s), slyA mutant (YS11068), and slyA mutant harboring pYS1109. Constitutively transcribed rpoD gene indicated that similar amounts of total RNA were used. The PCR products were separated in an agarose gel. Meanwhile, the protein levels of PhoP-HA, CorA-FLAG, and SlyA-FLAG were determined, respectively, using Western blot analysis in wild type (YS11591, YS11477, and YS10075), slyA mutant (YS11592, YS14099, and YS11068), and slyA mutant harboring pYS1109 (pslyA). Monoclonal anti-HA antibodies (Sigma) were used for HA-tagged protein and anti-FLAG M2 (Sigma) for FLAG-tagged protein. Constitutively produced CorA indicated that similar amounts of total protein were used. Percentage of the mRNA and protein amounts was calculated by formula: % mRNA or % protein (Pr) ϭ (ratio of individual phoP/rpoD or PhoP/CorA Ϭ ratio of wild type) ϫ 100. C, ␤-galactosidase activity from bacteria harboring pYS1100 (P phoP1 WT, with wildtype sequence), pYS1115 (P phoP1 up-6, with substituted actatt sequence), and pYS1244 (P phoP1 up-far, with substituted ttttctt sequence) was determined in wild type (14028s), phoP mutant (YS11590), phoP mutant harboring pYS1109 (pslyA), slyA mutant (YS11068), and slyA mutant harboring pYS1109 (pslyA). D, ␤-galactosidase activity from a chromosomal lacZ fusion in STM3595 and pcgL was determined in wild type (YS11620 and YS10382), phoP mutants (YS11754 and YS11743), phoP mutants harboring pslyA, slyA mutants (YS11621 and YS14071), and slyA mutants harboring pYS1109 (pslyA). Data in C and D correspond to three independent assays, and all graphed values are means Ϯ S.D. 2B). We studied the in vivo phoP transcription using a plasmid (pYS1100) carrying a lacZ transcriptional fusion to a phoP promoter fragment (P phoP1 ), in which lacZ expression was dependent on the PhoP/PhoQ system (17) (Fig. 2C). ␤-Galactosidase activity was 2.4-fold lower in a slyA mutant than in the wildtype strain, which further demonstrated SlyA feedback activating transcription of the phoPQ operon. The compatible plasmid harboring the slyA gene (pYS1109) complemented the deficient phenotype of the lacZ expression in the slyA mutant, but not in the phoP mutant (Fig. 2C), indicating that SlyA modulates transcription of the phoPQ operon through PhoP (i.e. autoregulation).
To see if changed PhoP levels could influence transcription of PhoP-activated genes, we constructed strains carrying a lacZ fusion in PhoP-activated chromosomal loci STM3595, pcgL, and many others demonstrated previously from different studies (29,33,34). Expression of lacZ from these strains was PhoPdependent because no ␤-galactosidase activity was detected when the phoP locus was mutated (Fig. 2D, and data not shown). Analysis of the lacZ expression indicates that SlyA facilitates expression of these selected genes in 0.01 mM Mg 2ϩ (Fig. 2D, and data not shown). Different from the results of pagC and ugtL activation (12,16), expression of these PhoPactivated genes decreased Ն2-fold, but were not turned off in slyA mutants ( Fig. 2D and data not  shown). This phenotype resulted solely from an absence of the SlyA protein because it could be complemented by pslyA plasmid (Fig. 2D). Similar to phoP transcription, this plasmid failed to rescue expression of these loci in a phoP mutant (Fig.  2C), indicating that SlyA-dependent activation requires a functional PhoP/PhoQ system. These results demonstrate a new mechanism that modulates regulatory activity of the PhoP/PhoQ system through SlyAmediated feedback transcriptional activation of the phoPQ operon.
SlyA Binds to the Promoter Region of the phoPQ Loci-SlyA binds to the promoter region of ugtL and pagC (12,16) to compete with H-NS, subsequently activating these type II PhoP-dependent genes (17). We hypothesized that SlyA mediates autoregulation of the phoPQ operon by interacting with its promoter. We carried out an EMSA using SlyA-FLAG protein, and found that this fusion protein alone gel-shifted the P phoP1 DNA fragment present in plasmid pYS1100 (Fig. 3A). Binding of the SlyA protein is further confirmed by a supershift assay using anti-FLAG M2 antibodies (Sigma) that inhibited SlyA-DNA interaction (Fig. 3A). This is probably because the antibody bound to the FLAG epitope and subsequently blocked the DNA recognition domain in the SlyA protein. Consistent with this notion, we found that phoP promoter DNA was shifted more by SlyA-FLAG in the presence of antibody when FLAG octapeptide (DYKDDDDK) was supplemented to reaction systems (Fig. 3A), in which this peptide competes for anti-FLAG antibody with SlyA-FLAG.
Next, we determined the DNA sequence in the phoP promoter recognized by SlyA protein using DNase I footprinting assays. We show that SlyA interacts with the AT-rich regions in the phoP promoter because the SlyA-FLAG protein protected the Ϫ106 to Ϫ101 and Ϫ90 to Ϫ72 region (numbering from the first ATG in the PhoP coding region) in the coding strand and the Ϫ109 to Ϫ102 and Ϫ88 to Ϫ76 region in the noncoding strand (Fig. 3B). The ACTATT sequence, which is located 6 nucleotides upstream of the PhoP box (Fig. 3C) and is identified carrying P phoP1 up-far. P1 is the PhoP-dependent transcription start site and the 3Ј end of P phoP1 fragment in pYS1100. The black box is the corresponding sequence of the phoP promoter from E. coli. The dots correspond to the conserved nucleotides in phoP promoters of Salmonella and E. coli. Numbering is from the A (as ϩ1) in the predicted phoP start codon (shown as uppercase letters). D, mRNA levels from E. coli were determined using RT-PCR in wild type (MC4100), and slyA mutant (YS14200) grown for 4 h in N medium (pH 7.4) containing 0.01 or 10 mM Mg 2ϩ . Percentage of the mRNA amount was calculated by formula: % mRNA ϭ (ratio of the individual phoP/rpoD Ϭ ratio of wild type from 0.01 mM Mg 2ϩ ) ϫ 100.
as an H-NS-binding site (i.e. up-6, see Ref. 17), overlaps the Ϫ90 to Ϫ72 sequence and resembles the ATTATT repeat (the SlyA box) from the pagC and ugtL promoters. 4 We compared lacZ expression from strains harboring pYS1100 (wild-type P phoP1 ) and pYS1100-derived plasmids with substitutions at the P phoP1 sequence. Surprisingly, ␤-galactosidase activity was similarly activated in wild type and slyA mutant harboring plasmid pup-6 with substitution at ACTATT (Fig. 2C). Because mutation of this sequence also abolished H-NS-mediated transcriptional repression in the phoPQ operon (17), we believe that SlyA is an effector that antagonizes the H-NS function in phoP regulation. Therefore, phenotype of a slyA mutant could be recessive when H-NS is absent (Fig. 2C). On the other hand, lacZ expression from wild type harboring another pYS1100-derived plasmid with a heptamer substitution at TTTTCTT within the Ϫ109 to Ϫ101 sequence (i.e. pup-far in Ref. 17), similar to pYS1100, is higher than that from the slyA mutant (Fig. 2C). These results indicate that binding of the SlyA protein to the region adjacent to the PhoP box in the phoP promoter is required for its transcriptional activation in wild-type bacteria. The promoter region corresponding to the Salmonella SlyA-binding site is a GC-rich DNA fragment in the E. coli phoP promoter (letters in white, Fig. 3C). When the in vivo binding of the SlyA-FLAG protein to the E. coli phoP promoter was approached using ChIP assays, no significant enrichment of the phoP DNA was observed (data not shown). These observations suggest that SlyA-facilitated phoP transcription occurs in Salmonella but not in E. coli. The mRNA levels of the phoP transcript were similar in wild type and slyA mutant (Fig. 3D), indicating that SlyA is unlikely to be integrated in the PhoP/PhoQ regulatory circuit in E. coli. Currently, we are elucidating the role of SlyA in fine-tuning of two-component signaling by systematically comparing the binding ability of SlyA and H-NS to promoter regions of the twocomponent systems in Salmonella and E. coli. 4 SlyA Facilitates the Interaction of the PhoP Protein to Its Own Promoter-To examine the in vivo binding of SlyA and PhoP to the phoP promoter by implementing ChIP assays, we constructed the following strains producing SlyA and PhoP from heterologous promoter P lac : slyA mutant and slyA phoP double mutant harboring plasmid pYS1109, which directs synthesis of SlyA-FLAG protein; phoP mutant and slyA phoP double mutant harboring plasmid pYS1177, which directs synthesis of PhoP-HA protein. We ruled out different binding abilities of these regulators to the phoP promoter caused by varied levels of the protein because bacterial cultures supplemented with 0.2 mM IPTG produced similar amounts of SlyA-FLAG or PhoP-HA proteins, regardless of their genetic background (Fig. 4, A and B). The phoP promoter DNA was enriched equally by the SlyA-FLAG protein from both slyA and slyA phoP mutants (Fig. 4A), indicating that SlyA was able to interact with the phoP promoter in vivo, and this binding did not require PhoP. However, more DNA was enriched by the PhoP-HA protein when SlyA is present (Fig. 4B), suggesting that SlyA facilitates PhoP in interacting with its own promoter. The ChIP assays were specific because there was no significant enrichment of DNA fragment when a control strain (i.e. untagged wild type) was tested (Fig. 4,  A and B).
We determined the mRNA level of the phoP transcript in strains harboring a chromosomal allele (pho-24) maintaining phosphorylated PhoP protein irrespective of Mg 2ϩ concentrations, in which expression of PhoP-activated genes was constitutively up-regulated (PhoP C phenotype, see Refs. 7,35). We assumed that the phosphorylated PhoP protein would keep binding to the phoP promoter before or while we performed our tests in such bacterial strains. Different from the result in wild type (Fig. 2B), the mRNA levels were similar in the pho-24 strain and its isogenic slyA mutant (Fig. 4C). This observation suggests that SlyA is not required for the regulatory activity of PhoP when it has been binding to its target promoters. We summarize that SlyA activates phoP regulation in a PhoP- . % DNA ϭ (individual density Ϭ individual input density) ϫ 100. % protein (Pr) ϭ (individual density Ϭ density in b) ϫ 100. Value "0" indicates an actual value was below 1%. B, in vivo PhoP binding to the phoP promoter was determined in phoP mutant (YS11590) harboring pYS1177 (pphoP) and slyA phoP double mutant (YS14047) harboring pYS1177 (pphoP). Wild-type 14028s was used in A and B as untagged strain. Input is total DNA, and IP is immunoprecipitated DNA. PCR amplification was performed for 26 cycles, and DNA fragments were separated in an agarose gel and visualized by ethidium bromide. % DNA and % protein were calculated as in B. C, mRNA levels were determined using RT-PCR in phoP C strain (pho-24, YS11249), and phoP C slyA mutant (YS11250). The PCR products were separated in an agarose gel. Percentage of the mRNA amount was calculated by formula: % mRNA ϭ (ratio of the individual phoP/rpoD Ϭ ratio of phoP C strain) ϫ 100. dependent manner, although it binds to the phoP promoter by interacting with the upstream region of the PhoP box in a PhoPindependent manner.
SlyA-mediated Signal Amplification Does Not Change Mg 2ϩ Responsiveness of the PhoP/PhoQ System-␤-Galactosidase activity derived from both P phoP1 -lacZ and STM3595-lacZ fusions in wild type was higher than that from the slyA mutant in any given Mg 2ϩ condition from 0.01 to 10 mM (Fig. 5A). A parameter, C1 ⁄ 2 , was then designed to represent the Mg 2ϩ concentration in the medium, which allows ␤-galactosidase activity expressed in a strain to remain over 50% of the maximum (the amount induced by the highest signal, i.e. grown in 0.01 mM Mg 2ϩ in this study). If SlyA could influence the PhoQ activity, we should observe varied values of C1 ⁄ 2 in wild type and the slyA mutant derived from curves representing the percentage of ␤-galactosidase activity versus the Mg 2ϩ concentration. We compared Mg 2ϩ -dependent expression with regard to the phoP and STM3595 genes by measuring the empirical values of the C1 ⁄ 2 from each strain. The results showed that C1 ⁄ 2 (WT) and C1 ⁄ 2 (slyA) derived from a PhoP-activated gene (phoP or STM3595) were similar (Fig. 5B), although actual values of the ␤-galactosidase activity were higher in wild type than in the slyA mutant (Fig. 5A). This suggests that SlyA does not modulate the Mg 2ϩ responsiveness of PhoQ, which would change the ratio of phosphorylated PhoP at any given signal level. Our result indicates that the presence of SlyA simply increases transcriptional level but does not influence Mg 2ϩ sensing of the PhoP/PhoQ system. Consistent with this notion, expression of several other PhoP-activated genes also gave similar C1 ⁄ 2 values from wild type and slyA mutant (data not shown).

DISCUSSION
We have identified a regulatory mechanism that is responsible for activating transcription of the PhoP/PhoQ two-component signaling system of S. enterica. We establish that the PhoP regulon is up-regulated by a positive feedback controlled by the PhoP-activated SlyA. This model is supported by the following data. (i) The level of PhoP and PhoP-activated gene products was reduced in slyA mutants (Table 3 and Fig. 1B). (ii) Transcription directed by PhoP-dependent promoters decreased in slyA mutants (Fig. 2, B-D). (iii) Interaction of SlyA to the phoP promoter was observed in vitro (Fig. 3,  A and B). (iv) The DNA-binding sites of SlyA and H-NS are overlapped (Fig. 3, B and C). (v) SlyA facilitates PhoP in binding to its own promoter in vivo (Fig. 4, A  and B). (vi) SlyA up-regulates the PhoP regulon in varied Mg 2ϩ conditions (Fig. 5A).
Transcription of a genetic locus may be more complicated than being turned on/off. It has been demonstrated that PhoPactivated PmrD functions as a regulator that enhances the PmrA/PmrB system in low Mg 2ϩ condition (9). On the other hand, the PmrA regulator could inhibit the synthesis of PmrD by negative feedback because ␤-galactosidase activity from a pmrD-lacZ fusion in a pmrA or pmrB mutant became about 2-fold higher than that in wild type (37). Apparently, fine-tuning of the transcription process was designed as a strategy to control the pmrD mRNA level and therefore to control the PmrD protein level. Here, we unraveled novel regulatory mechanisms underlying the transcriptional activation of the PhoP/ PhoQ system. We demonstrate that PhoP-activated SlyA functions as a positive feedback activator to facilitate transcription of the PhoP/PhoQ system, which therefore enhances PhoPactivated genes overall in Salmonella.
The SlyA protein seems to exert its effect as a positive regulator by antagonizing inhibitory action of the global regulator H-NS in the phoP expression. We proposed that two key components determine transcription of the phoPQ operon as follows: the PhoP protein that binds to the PhoP box and activates (i.e. autoregulates) phoP transcription; and the H-NS protein that interacts with sequence(s) adjacent to the PhoP box and represses this process (17). SlyA is likely to compete with H-NS because they share the same DNA sequence in the region adjacent to the PhoP box of the phoP promoter (Fig. 3C). The presence of this binding site may confer mutually exclusive binding of H-NS and PhoP to the phoP promoter. SlyA is specifically important for dynamic binding of PhoP because the mRNA level was similar if PhoP has been binding to the promoter region (Fig. 4C). SlyA should act on the phosphorylated instead of the unphosphorylated PhoP; otherwise, Mg 2ϩ -dependent expression of phoP and PhoP-dependent genes exhibited different C1 ⁄ 2 values in wild-type and the slyA mutant (Fig. 5B).
Very little information has been shown if or how a specific cellular response is involved in controlling multiple two-component systems. Our results (Table 3 and Fig. 1B) indicate that bacteria might integrate signaling networks, which respond to different environmental signals, by connecting them to a specific environmental condition. We propose a SlyA-mediated regulatory mechanism illustrated in Fig. 6. To initiate the regulatory cascade stimulated by a signal such as low Mg 2ϩ , acidic pH, or antimicrobial peptides, kinase activity of the sensor PhoQ is modulated, which mediates phosphorylation of the cognate regulator PhoP. The activated regulator then enhances SlyA protein at a transcriptional level, which up-regulates transcription of phoPQ, as well as ompR/envZ that responds to osmotic stress (19) and cpxRA that responds to extracellular stress (32) in enteric bacteria. Therefore, depletion of Mg 2ϩ , exposure to acidic pH, and antimicrobial peptides would generate stresses that require gene products from those loci controlled by other two-component systems. It remains to be investigated whether SlyA could activate other two-component systems by antagonizing the inhibitory effect of H-NS.